Mid-Term Project

Synopsis

Regork Grocery Chain invests copious amounts of time and money into creating coupons to attract customers and drive sales. However, I wanted to answer a key question that always remains with the use of coupons: Do coupons actually increase customer retention, or do they just attract one-time shoppers? By analyzing transaction data to examine patterns of coupon redemption, Regork can refine its coupon strategies. This will allow the company to target repeat buyers and enhance long-term profitability.

To answer this question, I separated my analysis into two sections, “Consumers” and “Products”. The “Consumers” section focuses on whether coupon users tend to return for future purchases, taking into account factors like transaction frequency and demographics. The “Products” section examines if certain products are more likely to influence customer retention through coupon use, exploring how specific items impact purchasing behavior. This two-pronged approach provides a clearer understanding of whether coupons foster loyalty or merely attract short-term shoppers.

Packages Required

The complete journey data set contains data over one year from a group of 2,469 frequent shopper households at a grocery store.

completejourney- Data sets

tidyverse- Contains many packages useful for data manipulation, exploration, and visualization

lubridate- Makes working with dates easier

RColorBrewer Used to color bar charts

Data Preperation

The complete journey data sets used for this analysis:

transactions <- get_transactions()
products <- products
demographics <- demographics

Exploratory Data Analysis

Consumers

In this section, I created several plots in order to better understand the purchasing habits of coupon users .vs. non-users.

*note: The retention metric I will be using is transaction frequency.

PART ONE

Analyzing transaction frequency for coupon users and non-users.

First, I started my analysis by comparing the transaction frequency of coupon users and non coupon users. I did this by identifying which household ids have and have not redeemed a coupon, then by calculating the transaction frequency for both users and non users, and finally by plotting the data using a box plot.

Looking at this plot, it is easy to see that transaction frequency is consistently higher for coupon users rather than non users, and that coupon users in general purchase more items from the store. However, this data does not take any outside factors that might impact transaction frequency like marital status, age, or income. So, in the next part, I compared transaction frequency for users and non-users while taking those factors into consideration.

PART TWO

Studying if consumer demographics are impacting customer retention.

In this part, I used the demographics data set to analyze whether martial status, age, or income impact transaction frequency for coupon users and non-users.

First, I started by studying the impact of marital status on transaction frequency. To do this, I merged the demographics and transactions data sets by household id. Then, I removed the NA values from the marital status column. Next, I calculated transaction frequency for each household and classified them as a coupon user or non-user. Then, I calculated the average transaction frequency for each type of household, and finally I created the bar chart.

There are several takeaways that can be taken away from this bar chart but I am going to highlight two.

There is a much higher retention rate for coupon users for both married and unmarried customers. This means that marital status is not directly impacting customer retention, coupon usage is.
Non-users are more likely to be unmarried, suggesting an opportunity for Regork to focus on engaging this demographic. By targeting unmarried non-users with tailored promotions with the goal of turning them into users, Regork is able to have a higher retention rate with unmarried customers.

Next I analyzed the impact of age on transaction frequency for coupon users and non-users. I started by converting the “age” column in the demographics data set into a factor with levels in chronological order. Next, I used some of the filtered data we used in the marital status plot to calculate the transaction frequency for each household and to classify as a user or non user. Then, I calculated the mean transaction frequency for each age range. Finally, I created the bar chart.

Similar to the martial status plot, I am going to highlight two key takeaways from this plot.

Coupon users show significantly higher retention rates across all age groups, indicating that customer retention is driven more by coupon usage than by age itself.
The 35-44 age group shows the largest gap in average transaction frequency between coupon users and non-users. This presents an opportunity for Regork to target non-users within this demographic to boost their retention rate.

To conclude this section and part, I created a plot to study the impact of income level on average transaction frequency for coupon users and non-users. I started by I started by calculating transaction frequency and classifying each household id as a user or non-user. Next I calculated the mean transaction frequency for each income level, and to wrap up, I created the bar chart.

For the first time, we see a demographic where non-coupon users have a higher average transaction frequency than coupon users. Key highlights include:

In the $200-249k income range, non-users show a higher average transaction frequency than coupon users by about 450 transactions. This is an unusual pattern, particularly since the retention rate for coupon users is significantly higher in the neighboring income brackets.
The biggest gap in retention occurs in the 150-174k income bracket, where coupon users show a significantly higher retention rate. This trend is also evident in the neighboring income brackets, 125-149k and 175-199k, which display similarly large differences in favor of coupon users.
Regork should focus on converting non-users with incomes below $124k into coupon users to boost retention within this demographic, as lower-income customers may be more price-sensitive and therefore more likely to respond positively to coupon incentives, leading to increased store visits and long-term loyalty.

Products

In the products section of this analysis, I used bar charts and line charts to examine whether specific products influence customer retention in relation to coupon usage.

PART ONE

Examining the top 10 products purchased with a coupon to investigate whether specific products influence customer retention regarding coupon usage.

To investigate the influence of products on customer retention through coupon usage, I started by identifying the top 10 products with the highest coupon redemption rates. I opted to focus on these top 10 product categories rather than selecting them randomly to avoid skewing the data, as several categories have no coupon purchases at all.

I executed the code by first merging the transactions and products data sets using the product ID, and I removed any missing values. Then, I filtered the transactions to include only those where a coupon was used. After isolating the top 10 products purchased with a coupon, I renamed certain product categories for clarity. Finally, I visualized the data in a bar chart to highlight these top products.

After taking time to digest this plot, what stood out to me most is that these are all products that are not “grocery store necessities” like milk and eggs, they are products that are purchased often because of their convenience. However, this plot does not clarify whether it is the coupons driving the sales of these products or if they would have been purchased regardless. In the next plot, I calculated the percentage of transactions in which these products are purchased with a coupon to gain further insights.

To begin my code for this plot, I used the merged data sets I created in the plot above to then calculate the percentage of transactions in which a coupon was used. Next, I once again renamed select categories to better fit in the plot, and to wrap up this code chunk, I created the bar chart.

My major takeaway from studying this plot is that the low percentage of times these products are purchased with a coupon indicates that it is not the products themselves driving coupon usage, but rather the allure of the coupons. This suggests that the effectiveness of coupons in influencing purchases may be more significant than the inherent appeal of the products.

Furthermore, I believe that my findings raise important questions about customer retention, the most pressing being: If customers are primarily motivated by coupons themselves rather than the products, does this lead to a reliance on discounts for repeat purchases? To answer this question, in the next part I created a line plot showing sales trends overtime for refrigerated dough and baked breads.

PART TWO

Assessing whether coupon usage for certain products leads to sustained increases in sales overtime, and/or a reliance on discounts for repeat purchases.

Understanding if coupon usage for select products leads to an increase in customer retention and sales will help Regork develop strategies that not only incentivize immediate sales but also foster long-term loyalty among customers. So in my final plots, I chose to analyze the long term sales trends of refrigerated dough and baked breads as these product categories are by percentage the first and tenth most purchased product categories with a coupon.

To begin, I started by merging my data sets and converting the dates from “transaction_timestamp” in the transactions data set into date form while extracting the month and year. Next, I once again renamed the the product categories to better fit in the plot. Then I created the new data set by filtering for all transactions with the product category of “baked breads” and “refrigerated dough”. Finally, I created both plots.

Two key takeaways from these plots:

It is clear that coupons are not driving the sales of these products, or that repeat product purchase is increasing due to the use of coupons. This conclusion is supported by the observation that sales of baked breads purchased with a coupon remain relatively consistent throughout the year. In contrast, sales of refrigerated dough with a coupon only rise during periods when overall sales for the product experience a spike.
It is crucial to recognize the data quality issues associated with this plot. A more effective analysis would involve displaying sales data over a 2 to 5-year period instead of just one year, allowing for a more comprehensive understanding of external factors, such as seasonal purchases during holidays. The noticeable spike in refrigerated dough sales toward the end of the year likely reflects consumer behavior related to holiday baking. By extending the time frame of the analysis, I would be better equipped to discern the overall trends and patterns in sales, rather than relying on a potentially skewed snapshot from a single year. This extended perspective would enable a more nuanced interpretation of how various factors influence sales and customer behavior over time.

Summary

After analyzing the data, do coupons actually increase customer retention, or do they just attract one-time shoppers?

CONCLUSIONS REGARDING CONSUMERS

The key takeaway from this section is that overall, coupons do increase customer retention. Simply put, coupon users demonstrate a higher transaction frequency (my measure of retention) compared to non-users, as evidenced by the box plot I created. However, answering my initial question isn’t entirely straightforward. Multiple factors such as consumer demographics, specifically marital status, age, and income, come into play and may influence purchasing behavior in complex ways.

After analyzing the bar charts that study the effect of these demographic factors on retention for both coupon users and non-users, the most important conclusion is that customer retention seems to be driven more by coupon usage than by demographic considerations alone. While demographics influence purchasing habits, the consistent retention among coupon users across various groups suggests that it’s the coupons, not the demographics, that primarily drive repeat purchases. This highlights the strong incentive power of coupons in fostering customer loyalty, regardless of other personal attributes.

CONCLUSIONS REGARDING PRODUCTS

Drawing conclusions about the products section of this analysis is not as easy as it was for the consumers section. To try and put my conclusion simply, it is not the discount of a specific product that is bringing customers to the store, it is the coupons themselves.

From the initial plot, it was not clear if the sales of the top 10 products purchased with a coupon are driven by coupons or if customers would have purchased those products regardless. To gain further insights, I analyzed the percentage of transactions where coupons were used for each product.The takeaway from this deeper analysis was that the low percentage of times these products were purchased with a coupon suggests that it’s the coupons themselves—not the products—that drive coupon usage.

To explore whether coupon usage leads to sustained increases in sales, or if they create a reliance on discounts for repeat purchase, I created line plots to track the long-term sales trends of baked breads and refrigerated dough. From these plots, I drew the conclusion that coupon usage for certain products does not lead to increases in sales overtime, and that coupons do not create a reliance on discounts for repeat purchase.

FINAL CONCLUSION

After thoroughly analyzing the data, it’s evident that coupons do play a significant role in increasing customer retention, though the full story is more complex. In terms of consumer behavior, coupon users consistently show higher transaction frequency compared to non-users, suggesting that coupons do foster repeat purchases. While demographic factors such as marital status, age, and income also influence shopping patterns, the data shows that retention is primarily driven by coupon usage, with demographics playing a secondary role.

On the product side, the analysis indicates that it’s not specific products driving coupon use, but rather the appeal of coupons themselves. Products purchased with coupons are typically convenience items, and the low percentage of transactions involving coupons suggests that customers are motivated by discounts more than by any inherent attraction to these items. Line plots tracking long-term sales trends for baked breads and refrigerated dough confirm that coupons do not lead to sustained sales increases or create a dependence on discounts for repeat purchases.

In conclusion, while coupons are effective in driving customer retention and repeat purchases, their influence is largely independent of the products themselves. They act as a powerful incentive, encouraging loyalty more through the promise of savings than through product preferences or demographic influences.